Why have Latin American central bankers departed from the conventional inflation targeting script in terms of exchange rate management and reserves accumulation? The answer may be that the applicability of the conventional script to Latin American economies is believed to be only partial. It turns out that good reasons for such a belief can be found in recent academic work. In this section we review some of those reasons, and we also ask whether they in fact warrant observed policies.
4.1. Balance Sheet Effects and Dollarization
The New Keynesian open economy models underlying the conventional IT wisdom are built upon the assumption that an exchange rate depreciation is expansionary. This is because, in those models, the main channel through which a depreciation affects aggregate demand is by changing the relative price of domestic vis a vis foreign goods and by reducing domestic interest rates.
The view that depreciations are expansionary, however, has been long at odds with empirical evidence, and has come under substantial challenge following the crises in Mexico 1995 and Asia 1997-98, the Russian 1998 default, and the Brazilian 1999 devaluation. Those episodes were associated with widespread financial contagion and exchange rate depreciation in many emerging economies. And in several cases it was observed that a weaker currency impaired the balance sheets of debtor firms, reducing the access of those firms to credit and, as a consequence, restricting investment demand. The balance sheet impact of a depreciating
exchange rate was greater for those agents that had taken debts in foreign currency (typically the US dollar) when their revenues were given in domestic currency.
These observations have motivated a substantial body of research focusing on the links between currency mismatches, balance sheet effects, and monetary and exchange rate policy. A main finding of these research efforts is that it is not too hard to identify circumstances in which exchange rate depreciations are, in fact, contractionary instead of expansionary. In particular, if corporate firms have dollar debts but their assets or revenues are determined by the value of the domestic currency, a fall in the exchange rate reduces their net worth. The fall in net worth, in turn, can easily lead to reduced credit. Why? If financial markets are characterized by asymmetric information, imperfect contract enforcement, or other frictions, lenders may require borrowing firms to secure any loans with collateral funds. But the availability of the latter is directly related to each firm’s net worth.
So Leiderman, Maino, and Parrado (2006) write that, in an economy characterized by currency mismatches,
it is plausible to argue that balance sheet effects will give rise to contractionary devaluations and induce financial stress...In turn, the potentially adverse effect of large exchange rate devaluations is likely to induce fear of floating by the authorities and require that they closely target the exchange rate, even when the underlying shocks are transitory.
This reasoning implies that the incidence of balance sheet effects depends on the degree of dollarization, as well as on the severity of the imperfections of the financial system. Central bankers in highly dollarized economies will naturally be concerned with the potential impact of an exchange rate depreciation on corporate balance sheets, and their concern will be greater the less developed the financial system.
considerations in Peru, which is the most dollarized of the countries under review and has a relatively small financial sector. According to Armas and Grippa (2006), the Peruvian Banco Central de Reserva "limits balance sheet effects by moderating the volatility of the exchange rate." Armas and Grippa also mention that, in fact, the Peruvian monetary strategy explicitly allows for switching the policy instrument from the normal overnight interbank interest rate to a monetary aggregate precisely because of potential balance sheet effects.
While strong balance sheet effects raise concern about contractionary devaluation, it is not obvious that the proper monetary policy response is to stabilize the exchange rate. This is because, when balance sheet effects are present, they usually depend on the real exchange rate, not the nominal exchange rate. Monetary policy can stabilize the latter, but cannot prevent a real exchange rate adjustment if one is required. This point was developed by Cespedes, Chang, and Velasco (2004) which studied a New Keynesian open economy model with balance sheet effects. In their model, an inflation targeting policy that stabilizes the price of domestically produced goods is socially optimal and, in particular, it dominates fixed exchange rates. The reason is that fixing nominal exchange rates does not prevent real exchange rate movements, and it is the latter which cause strong balance sheet effects.31 That the presence of balance sheet effects leaves fixed exchange rates inferior to inflation targeting is also found by Devereux, Lane, and Xu (2004), Gertler, Gilchrist and Natalucci (2003), and many others.
This is not to say that central bankers should not pay attention to the balance sheet effects of exchange rate movements, but to stress that we are far from understanding how
31To see why this happens in the Cespedes- Chang-Velasco model, let P Y denote the revenues of domestic
firms, where Y is production and P the domestic currency price of home produce. If S is the exchange rate (domestic currency per unit of foreign currency), firms’ revenues in foreign currency are (P/S)Y. Hence, if the firms have debts in foreign currency, the value of the latter relative to revenues increases if S/P increases, that is, if there is a real depreciation. A fixed exchange rate means that S is fixed, but S/P can (and does, in the model) adjust in response to shocks.
the setting of their policy instruments should be in view of those effects. Perhaps the best response, instead of defending the exchange rate against depreciation, will turn out to be to foster the de-dollarization of the financial system (as the Peruvian Banco Central is actually doing), as well as to work on the development of more efficient and deeper capital markets.
4.2. Crises, International Liquidity, and Reserves Accumulation
A somewhat different lesson from the period of emerging markets crises of the last decade was that international liquidity is a key factor on the generation of crises. Following Chang and Velasco (2000) we will say that a country has an international liquidity shortage if its potential short term liabilities in international currency exceed the liquidation value, also in hard currency, of its assets. A crisis can then occur if the country’s creditors panic and demand immediate payment of their short term claims. In that case, liquidating the country’s assets is insufficient to honor the claims, leading to bankruptcy.
The concept of international liquidity is useful in understanding crises for a variety of reasons. First, it focuses squarely on financial contracts and institutions, as opposed to other theories of open economy crises which assigned a prime role to inconsistent macroeconomic policy. This is significant because the emerging markets crises of the 1990s looked a lot like crises of the financial system, while the evidence of a flawed macroeconomic policy was decidedly mixed.
Second, a country with an international liquidity problem may or may not experience a crisis. Whether a crisis materializes depends on market expectations, since it is only if creditors attempt to convert their potential short term claims into actual claims that the country goes bankrupt. If they do not, the country may remain in a favorable no crisis equilibrium. This is consistent with at least three observations: that several crises were almost completely unexpected by the markets; that market expectations and psychology were assigned a prime role during crisis periods; and that there was widespread contagion,
that is, financial panics affected countries that had little in common with the original crisis countries.
Third, the theoretical links between international liquidity and crises were consistent with a variety of evidence. Most convincingly, econometric studies found that the probability of a crisis was increasing on the ratio of short term foreign debt to reserves (see Rodrik and Velasco 2000, for instance). But the short term debt/reserves ratio is an obvious indicator of international liquidity.
As the view that international liquidity was a prime factor behind crises, emerging country governments have started taking active steps towards reducing vulnerabilities in this regard. One concrete instance is the effort to reduce reliance on short term debt denominated in foreign currency, with many governments now issuing domestic currency denominated debt. In a complementary fashion, emerging country governments have stepped up efforts to build up international reserve levels, which can be sufficient to deal with the sudden demands for international liquidity that characterized recent crises.
This is why the four central banks reviewed here have explicitly stated that the accumu- lation of international reserves is one of the tasks to be performed in an inflation targeting framework. As mentioned earlier, Colombia’s Banco de la Republica notes that foreign ex- change intervention is to be allowed "to maintain an adequate level of international reserves that reduce the vulnerability of the economy to foreign shocks." For Peru, Armas and Grippa (2006) write:
If a bank run on foreign currency denominated deposits occurs, a high level of net international reserves acts as a buffer stock for supporting the financial system. Moreover, the availability of international reserves may in itself be an insurance against negative shocks...Since the beginning of 2003, the BCRP has purchased around US$ 4.5 billion (as of 31 March 2005) in the foreign exchange
market...Thus, international reserves exceeded US$13.5 billion in March 2005, the highest level ever recorded. This is an important buffer stock against any shock, considering that it is more than twice the stock of the due-in-one-year external debt.
The Central Bank of Chile (2007) states:
[The stock of international reserves] also reduce the probability of liquidity perturbations and make it possible to face exceptional situations of loss of access to international financial markets, minimizing the probability of problems in balance of payments...the Banco Central de Chile monitors the adequate level of international reserves as well as its composition. 32
And, as we stressed, the current buildup of international reserves in Brazil is justified on the basis that a large stock of reserves provides a "vaccine against international turbulence." Overall, then, current efforts at accumulating international reserves are consistent with the view that international liquidity matters for crises. Yet one can identify several issues that deserve further study.
First, there is nothing in the theory that says that hard currency "war chests" should be built by the central bank, as opposed to a distinct government agency similar, say, to the U.S. Treasury’s Exchange Stabilization Fund. And there may be good reasons to keep the war chests separate from the central bank. One could argue, for example, that the current arrangements, by confounding policies to achieve the inflation targets with attempts at accumulating reserves, undermine the stated objectives of transparency and accountability that an inflation targeting regime should aim at. 33
32La Politica Monetaria de Chile en el Marco de Metas de Inflación, p. 20
33Notably, and perhaps in response to this concern, Chile set up an Economic Stabilization Fund as a
separate entity from the Central Bank, which presumably has allowed the latter to focus on the attainment of its inflation target.
Second, the accumulation of foreign exchange reserves is costly, and perhaps inefficiently so. International reserve assets pay little or no interest, so holding them implies an op- portunity cost approximately equal to the interest foregone by not holding interest bearing assets or by financing reserves with public debt. If purchasing foreign exchange is possible because of the issuance of public debt, the cost of the "turbulence vaccine" is increased fiscal expenditures later on, to finance the interest on the debt. More generally, one could ask why it is necessary to hold international reserves to match short term debt, as opposed to using reserves to prepay the debt.
Existing theories also imply that reserves accumulation is not the only way to prevent liquidity driven crises. And some of the alternatives are less costly. For example, as stressed by Chang and Velasco (2000) and others, a credible announcement that payments of the short term debt would be suspended in case of a panic could do the job at no cost. Alternatively, liquidity driven crises would be ruled out for a country that could ensure enough access to international liquidity in a crisis; one way this could be accomplished would be to set up an international line of credit akin to the Contingent Credit Lines offered by the IMF.
Third, the international liquidity argument provides a justification for the accumulation of reserves, but only up to a point. Once short term potential liabilities in hard currency are covered by international reserves, there is no further gain from additional reserves. This point seems to have been reached in several Latin countries, such as Brazil and Peru, yet reserves accumulation continues.
These considerations have motivated an ongoing and spirited debate on the costs of holding large amounts of international reserves, as well as the relative size of the chest of reserves that is necessary to ameliorate the possibility of crises. Rodrik (2005) argued that the annual cost of the current reserve buildup by emerging economies is around one percent of their GDP, although Lipschitz, Messmacher and Mourmouras (2006) dispute the accuracy of Rodrik’s calculations.
As for the size of war chests, two rules of thumb have emerged. The first is the so called Greenspan-Guidotti rule, which prescribes that emerging economies should hold reserves of the same magnitude as their short term foreign debt. The second is due to Jeanne and Ranciere (2006), and suggests that emerging economies should hold no less than ten percent of their annual GDP as reserves.
Whatever one thinks of these rules of thumb, it is instructive to ask how they apply to the countries we have reviewed. The ratio of Brazil’s reserves to GDP at the end of June 2007 stood at 14.3 percent, and hence well above the Jeanne-Ranciere ten percent guideline. Likewise, Brazil’s net reserves exceeded its short term external debt by more than US$ 124 billion.34 In Peru, international reserves stood at US$ 18.5 billion at the end of March 2007,
almost six times the amount needed to service the short term external debt of US$ 3 billion. In addition, the ratio of reserves to 2006 GDP stood at 19.2 percent.
It is then clear that it is no longer easy to justify the current efforts at reserves accumula- tion on the basis of international liquidity considerations. At this point, there is widespread agreement that Latin countries hold enough reserves to forestall any conceivable financial panic.
4.3. Nonfundamental exchange rate volatility
As we have seen, Latin American central banks have frequently justified a monetary policy response to exchange rates whenever the latter exhibits movements that are considered ex- cessive. Presumably, "excessive" is taken to mean that those movements were not warranted by the economy’s fundamentals. The academic literature has tried to model such excessive volatility in at least two related ways: multiple equilibria and bubbles.
One way to think about excessive exchange rate volatility is to admit the possibility that an economy has several self fulfilling equilibria, some of which display larger exchange rate
movements than others. Exchange rate volatility can then be excessive in the sense that, with the fundamentals of the economy being given, the economy could move to an alternative equilibrium with less volatility, presumably if only market expectations could coordinate on the more stable equilibrium.
One policy implication is that, if policymakers believe that the economy is stuck in an equilibrium with high exchange rate volatility but a better, less volatile equilibrium exists, then it may be advisable to intervene in the foreign exchange market to smooth exchange rate fluctuations, in the hope of convincing market participants to settle in the less volatile equilibrium.
While this chain of reasoning sounds plausible, unfortunately no satisfactory theory exists of how a particular equilibrium is selected when there may be many and, hence, of how foreign exchange intervention may work to switch from a bad to a good equilibrium. Hence, the practical approach to this issue remains at the level of art rather than science.
This is acknowledged, for example, by De Gregorio, Tokman, and Valdes (2005) in the case of Chile:
The CB has made public the view that interventions could be warranted in cases when there is an overreaction of the exchange rate and that this overre- action could be damaging for the economy...The CB has explicitly recognized that detecting an exchange rate overreaction or overshooting is not an easy task. Even in cases where there is no full certainty that there is an overreaction, it may be advisable to intervene. In some occasions such interventions may be ineffec- tive, but the cost of [not] intervening outweighs the expected costs of unjustified turmoil...At the end, determining exceptional circumstances or whether the ex- change rate is overreacting is a judgmental call of the CB Board. However, it is an informed call...
A second way to model excessive exchange rate volatility is by allowing for the existence of bubbles in the exchange rate. An exchange rate bubble is a deviation of the exchange rate from the present discounted value of its expected fundamentals. The bubble can be rational in the short run because of the expectation of an even larger bubble tomorrow, and so on.35
Bubbles have proven to be more amenable to modeling than multiple equilibria. For closed economies, an important paper is Bernanke and Gertler (2000), which discussed the role of asset price bubbles in a model with balance sheet effects and inflation targeting. In such a model, asset bubbles and balance sheets interact, since the value of corporate net worth may depend on asset prices inclusive of bubbles. In such a case, a positive bubble (i.e. an episode of "irrational exuberance") inflates net worth and the value of collateral, expanding access to credit, and boosting investment and aggregate demand. Hence a positive asset bubble is expansionary and contributes to inflation. Surprisingly, however, Bernanke and Gertler argue in favor of an inflation targeting regime in which the monetary instrument